@aglyn/plugins-ai 1.0.0-beta.270 → 1.0.0-beta.272

package/src/lib/site-import/site-import-html.js

@@ -0,0 +1,462 @@
+/**
+ * @license
+ * Copyright 2026 Aglyn LLC
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *   http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */ /*==========================================
+ * WHAT A CRAWLED PAGE SAYS (AGL-3974, parent AGL-3973).
+ *
+ * "Bring my site" reads a person's existing pages for their words, their
+ * structure and their pictures. No server HTML parser is a dependency of any
+ * lib, and a page's text in order is a small question, so this is a
+ * tokenizer: tags and text runs, with the few elements that matter tracked
+ * and everything else read through.
+ *
+ * What never survives, before a single tag is walked:
+ * - scripts, styles, iframes, embeds, objects, SVG, canvas, media players,
+ *   templates and comments, content and all;
+ * - any form with a card, bank or payment field, whole;
+ * - and a page with a password field is not read at all: it is refused and
+ *   reported (`loginForm`), because a sign-in page is exactly what a
+ *   phishing clone wants and nothing a site import needs.
+ *
+ * The output is DATA. Text from a page reaches a model only fenced as
+ * untrusted (`fenceUntrustedText`), never as instructions.
+ *=========================================*/ /** One run of a page's content, in the order the page draws it. */ import { _ as _extends } from "@swc/helpers/_/_extends";
+/** Ceilings that keep one odd page from becoming the whole inventory. */ export const SITE_IMPORT_PAGE_LIMITS = {
+    blocks: 400,
+    blockChars: 2000,
+    images: 100,
+    links: 300,
+    navLinks: 60,
+    forms: 10,
+    fieldsPerForm: 30
+};
+/** Elements removed with everything inside them. */ const REMOVED_WITH_CONTENT = [
+    'script',
+    'style',
+    'noscript',
+    'template',
+    'iframe',
+    'frame',
+    'frameset',
+    'object',
+    'applet',
+    'svg',
+    'math',
+    'canvas',
+    'video',
+    'audio'
+];
+/** Form controls, removed with their content once the forms are summarized. */ const FORM_CONTROLS = [
+    'textarea',
+    'select',
+    'button'
+];
+/** Void elements removed wherever they appear. */ const REMOVED_VOID = /<(?:embed|param|source|track|link|base)\b(?:[^>"']|"[^"]*"|'[^']*')*>/gi;
+const PASSWORD_INPUT = /<input\b(?:[^>"']|"[^"]*"|'[^']*')*\btype\s*=\s*["']?password\b/i;
+/**
+ * A field that takes money or a bank detail: an autocomplete hint for a
+ * card, or a name or id that says card, CVC, IBAN, routing or payment.
+ */ const PAYMENT_FIELD = /<(?:input|select)\b(?:[^>"']|"[^"]*"|'[^']*')*?(?:\bautocomplete\s*=\s*["']?\s*cc-|\b(?:name|id|data-stripe|data-braintree-name)\s*=\s*["']?[^"'\s>]*(?:card|cc[-_]?(?:num|number|exp|csc|cvv|cvc)|cvv|cvc|csc|iban|routing|account[-_]?number|payment|billing[-_]?card))/i;
+const BLOCK_TAGS = {
+    p: 'paragraph',
+    li: 'list-item',
+    dt: 'paragraph',
+    dd: 'paragraph',
+    td: 'paragraph',
+    th: 'paragraph',
+    address: 'paragraph',
+    pre: 'paragraph',
+    figcaption: 'paragraph',
+    caption: 'paragraph',
+    label: 'paragraph',
+    blockquote: 'quote'
+};
+/** Tags that end a run of loose text without being a block themselves. */ const BREAK_TAGS = new Set([
+    'div',
+    'section',
+    'article',
+    'main',
+    'aside',
+    'header',
+    'footer',
+    'nav',
+    'ul',
+    'ol',
+    'dl',
+    'table',
+    'tr',
+    'tbody',
+    'thead',
+    'figure',
+    'form',
+    'hr',
+    'br',
+    'body'
+]);
+const NAMED_ENTITIES = {
+    amp: '&',
+    lt: '<',
+    gt: '>',
+    quot: '"',
+    apos: "'",
+    nbsp: ' ',
+    rsquo: '’',
+    lsquo: '‘',
+    rdquo: '”',
+    ldquo: '“',
+    mdash: '—',
+    ndash: '–',
+    hellip: '…',
+    copy: '©',
+    reg: '®',
+    trade: '™',
+    middot: '·',
+    bull: '•',
+    eacute: 'é',
+    egrave: 'è',
+    aacute: 'á',
+    ntilde: 'ñ',
+    uuml: 'ü',
+    ouml: 'ö',
+    auml: 'ä',
+    deg: '°',
+    times: '×'
+};
+/** Character references to characters; an unknown name stays as written. */ export function decodeHtmlEntities(text) {
+    return text.replace(/&(#x[0-9a-f]+|#\d+|[a-z][a-z0-9]*);/gi, (whole, body)=>{
+        if (body[0] === '#') {
+            const code = body[1] === 'x' || body[1] === 'X' ? parseInt(body.slice(2), 16) : parseInt(body.slice(1), 10);
+            if (!Number.isFinite(code) || code <= 0 || code > 0x10ffff) return '';
+            if (code < 32 && code !== 9 && code !== 10 && code !== 13) return '';
+            return String.fromCodePoint(code);
+        }
+        const named = NAMED_ENTITIES[body.toLowerCase()];
+        return named != null ? named : whole;
+    });
+}
+const ATTRIBUTE = /([^\s=/>"']+)(?:\s*=\s*(?:"([^"]*)"|'([^']*)'|([^\s>"']+)))?/g;
+/** A tag's attributes, names lowercased, values decoded. */ export function parseAttributes(source) {
+    const attributes = {};
+    for (const match of source.matchAll(ATTRIBUTE)){
+        var _ref, _ref1, _match_;
+        const name = match[1].toLowerCase();
+        if (name in attributes) continue;
+        attributes[name] = decodeHtmlEntities((_ref = (_ref1 = (_match_ = match[2]) != null ? _match_ : match[3]) != null ? _ref1 : match[4]) != null ? _ref : '');
+    }
+    return attributes;
+}
+const TAG = /<(\/?)([a-zA-Z][\w:-]*)((?:[^>"']|"[^"]*"|'[^']*')*)>/g;
+/** Every `<meta>` tag's attributes. */ function metaTags(html) {
+    const out = [];
+    for (const match of html.matchAll(/<meta\b((?:[^>"']|"[^"]*"|'[^']*')*)>/gi)){
+        out.push(parseAttributes(match[1]));
+    }
+    return out;
+}
+/** The `content` of every `<meta name="…">` with this name. */ export function readMetaValues(html, name) {
+    const wanted = name.toLowerCase();
+    return metaTags(html).filter((attributes)=>{
+        var _attributes_name;
+        return ((_attributes_name = attributes['name']) != null ? _attributes_name : '').toLowerCase() === wanted;
+    }).map((attributes)=>{
+        var _attributes_content;
+        return ((_attributes_content = attributes['content']) != null ? _attributes_content : '').trim();
+    }).filter(Boolean);
+}
+function collapse(text) {
+    return text.replace(/\s+/g, ' ').trim();
+}
+function clip(text, max) {
+    return text.length > max ? `${text.slice(0, max - 1).trimEnd()}…` : text;
+}
+function removeElements(html, tags) {
+    let out = html;
+    for (const tag of tags){
+        const paired = new RegExp(`<${tag}\\b(?:[^>"']|"[^"]*"|'[^']*')*>[\\s\\S]*?<\\/${tag}\\s*>`, 'gi');
+        out = out.replace(paired, ' ');
+        // One left open runs to the end of the document, as a browser reads it.
+        const unclosed = new RegExp(`<${tag}\\b(?:[^>"']|"[^"]*"|'[^']*')*>[\\s\\S]*$`, 'i');
+        out = out.replace(unclosed, ' ');
+    }
+    return out;
+}
+/** Forms with a payment field go whole; the rest are summarized, then go. */ function takeForms(html) {
+    const kept = [];
+    let removed = 0;
+    const out = html.replace(/<form\b(?:[^>"']|"[^"]*"|'[^']*')*>[\s\S]*?(?:<\/form\s*>|$)/gi, (form)=>{
+        if (PAYMENT_FIELD.test(form)) {
+            removed += 1;
+            return ' ';
+        }
+        if (kept.length < SITE_IMPORT_PAGE_LIMITS.forms) {
+            const fields = [];
+            for (const match of form.matchAll(/<(input|textarea|select)\b((?:[^>"']|"[^"]*"|'[^']*')*)>/gi)){
+                const attributes = parseAttributes(match[2]);
+                const type = match[1].toLowerCase() === 'input' ? (attributes['type'] || 'text').toLowerCase() : match[1].toLowerCase();
+                if (type === 'hidden' || type === 'submit' || type === 'button' || type === 'image') continue;
+                fields.push({
+                    type,
+                    name: clip(attributes['name'] || attributes['id'] || '', 80)
+                });
+                if (fields.length >= SITE_IMPORT_PAGE_LIMITS.fieldsPerForm) break;
+            }
+            kept.push({
+                fields
+            });
+        }
+        return ' ';
+    });
+    return {
+        html: out,
+        kept,
+        removed
+    };
+}
+/** `raw` against the page, as an absolute http(s) address without a fragment. */ export function resolveSiteUrl(raw, base) {
+    const value = String(raw != null ? raw : '').trim();
+    if (!value || /^(?:javascript|data|mailto|tel|sms|blob|about):/i.test(value)) return null;
+    try {
+        const url = new URL(value, base);
+        if (url.protocol !== 'https:' && url.protocol !== 'http:') return null;
+        url.hash = '';
+        return url;
+    } catch (unused) {
+        return null;
+    }
+}
+/** The first address a `srcset` offers. */ function firstSrcset(srcset) {
+    var _String_split_;
+    const first = (_String_split_ = String(srcset != null ? srcset : '').split(',')[0]) == null ? void 0 : _String_split_.trim().split(/\s+/)[0];
+    return first || undefined;
+}
+/**
+ * The words, structure, pictures and links of one page.
+ *
+ * `isSameSite(url)` says whether an address is on the site being imported;
+ * only those become `links`, and images are marked by it.
+ */ export function extractSiteImportPage(html, pageUrl, isSameSite) {
+    var _ref;
+    var _parseAttributes_lang;
+    const empty = {
+        title: '',
+        description: '',
+        lang: null,
+        canonical: null,
+        blocks: [],
+        images: [],
+        navLinks: [],
+        links: [],
+        forms: {
+            kept: [],
+            removed: 0
+        },
+        loginForm: false
+    };
+    let source = String(html != null ? html : '').replace(/<!--[\s\S]*?(?:-->|$)/g, ' ').replace(/<!\[CDATA\[[\s\S]*?\]\]>/g, ' ');
+    if (PASSWORD_INPUT.test(source)) return _extends({}, empty, {
+        loginForm: true
+    });
+    source = removeElements(source, REMOVED_WITH_CONTENT).replace(REMOVED_VOID, ' ');
+    // Forms next, so a field list is read before the controls go.
+    const forms = takeForms(source);
+    source = removeElements(forms.html, FORM_CONTROLS);
+    const metas = metaTags(source);
+    const metaContent = (key, attribute)=>{
+        var _ref;
+        var _metas_find;
+        return collapse((_ref = (_metas_find = metas.find((m)=>{
+            var _m_attribute;
+            return ((_m_attribute = m[attribute]) != null ? _m_attribute : '').toLowerCase() === key;
+        })) == null ? void 0 : _metas_find['content']) != null ? _ref : '');
+    };
+    const titleMatch = /<title\b[^>]*>([\s\S]*?)<\/title\s*>/i.exec(source);
+    const title = clip(collapse(decodeHtmlEntities((_ref = titleMatch == null ? void 0 : titleMatch[1]) != null ? _ref : '')) || metaContent('og:title', 'property'), 300);
+    const description = clip(metaContent('description', 'name') || metaContent('og:description', 'property'), 1000);
+    const htmlTag = /<html\b((?:[^>"']|"[^"]*"|'[^']*')*)>/i.exec(source);
+    const lang = htmlTag ? ((_parseAttributes_lang = parseAttributes(htmlTag[1])['lang']) == null ? void 0 : _parseAttributes_lang.trim()) || null : null;
+    let canonical = null;
+    for (const match of String(html != null ? html : '').matchAll(/<link\b((?:[^>"']|"[^"]*"|'[^']*')*)>/gi)){
+        var _attributes_rel;
+        const attributes = parseAttributes(match[1]);
+        if (((_attributes_rel = attributes['rel']) != null ? _attributes_rel : '').toLowerCase().split(/\s+/).includes('canonical')) {
+            var _ref1;
+            var _resolveSiteUrl;
+            canonical = (_ref1 = (_resolveSiteUrl = resolveSiteUrl(attributes['href'], pageUrl)) == null ? void 0 : _resolveSiteUrl.toString()) != null ? _ref1 : null;
+            break;
+        }
+    }
+    const bodyStart = source.search(/<body\b/i);
+    const body = bodyStart >= 0 ? source.slice(bodyStart) : source.replace(/<head\b[\s\S]*?<\/head\s*>/i, ' ');
+    const hasNav = /<nav\b/i.test(body);
+    const blocks = [];
+    const images = [];
+    const imageSeen = new Set();
+    const navLinks = [];
+    const links = [];
+    const linkSeen = new Set();
+    let navDepth = 0;
+    let headerDepth = 0;
+    let footerDepth = 0;
+    let current = null;
+    let buffer = '';
+    let anchor = null;
+    const region = ()=>footerDepth > 0 ? 'footer' : headerDepth > 0 ? 'header' : 'main';
+    // A site without a <nav> keeps its menu in the header; its links are the nav.
+    const inNav = ()=>navDepth > 0 || !hasNav && headerDepth > 0;
+    const flush = ()=>{
+        var _ref;
+        const text = clip(collapse(decodeHtmlEntities(buffer)), SITE_IMPORT_PAGE_LIMITS.blockChars);
+        buffer = '';
+        const kind = (_ref = current == null ? void 0 : current.kind) != null ? _ref : 'paragraph';
+        const level = current == null ? void 0 : current.level;
+        current = null;
+        if (!text || navDepth > 0 || blocks.length >= SITE_IMPORT_PAGE_LIMITS.blocks) return;
+        const last = blocks[blocks.length - 1];
+        if (last && last.kind !== 'image' && last.text === text) return;
+        if (kind === 'heading') {
+            blocks.push({
+                kind,
+                level: level != null ? level : 2,
+                text,
+                region: region()
+            });
+        } else {
+            blocks.push({
+                kind,
+                text,
+                region: region()
+            });
+        }
+    };
+    for (const match of body.matchAll(new RegExp(`${TAG.source}|([^<]+)|<`, 'g'))){
+        var _match_;
+        const text = match[4];
+        if (text !== undefined || match[2] === undefined) {
+            const run = text != null ? text : '<';
+            buffer += run;
+            if (anchor) anchor.text += run;
+            continue;
+        }
+        const closing = match[1] === '/';
+        const tag = match[2].toLowerCase();
+        const attributes = closing ? {} : parseAttributes((_match_ = match[3]) != null ? _match_ : '');
+        if (tag === 'a') {
+            if (closing) {
+                if (anchor) {
+                    const label = clip(collapse(decodeHtmlEntities(anchor.text)), 120);
+                    if (inNav() && label && navLinks.length < SITE_IMPORT_PAGE_LIMITS.navLinks) {
+                        navLinks.push({
+                            text: label,
+                            url: anchor.url.toString()
+                        });
+                    }
+                    anchor = null;
+                }
+                continue;
+            }
+            const url = resolveSiteUrl(attributes['href'], pageUrl);
+            anchor = url ? {
+                url,
+                text: ''
+            } : null;
+            if (url && isSameSite(url) && links.length < SITE_IMPORT_PAGE_LIMITS.links) {
+                const key = url.toString();
+                if (!linkSeen.has(key)) {
+                    linkSeen.add(key);
+                    links.push(key);
+                }
+            }
+            continue;
+        }
+        if (tag === 'img' && !closing) {
+            const url = resolveSiteUrl(attributes['src'] || attributes['data-src'] || firstSrcset(attributes['srcset'] || attributes['data-srcset']), pageUrl);
+            if (url && images.length < SITE_IMPORT_PAGE_LIMITS.images) {
+                var _attributes_alt;
+                const key = url.toString();
+                const alt = clip(collapse((_attributes_alt = attributes['alt']) != null ? _attributes_alt : ''), 300);
+                if (!imageSeen.has(key)) {
+                    imageSeen.add(key);
+                    images.push({
+                        url: key,
+                        alt,
+                        sameSite: isSameSite(url)
+                    });
+                    if (navDepth === 0) {
+                        flush();
+                        if (blocks.length < SITE_IMPORT_PAGE_LIMITS.blocks) {
+                            blocks.push({
+                                kind: 'image',
+                                url: key,
+                                alt,
+                                region: region()
+                            });
+                        }
+                    }
+                }
+            }
+            continue;
+        }
+        if (tag === 'nav' || tag === 'header' || tag === 'footer') {
+            flush();
+            const step = closing ? -1 : 1;
+            if (tag === 'nav') navDepth = Math.max(0, navDepth + step);
+            if (tag === 'header') headerDepth = Math.max(0, headerDepth + step);
+            if (tag === 'footer') footerDepth = Math.max(0, footerDepth + step);
+            continue;
+        }
+        const heading = /^h([1-6])$/.exec(tag);
+        if (heading || BLOCK_TAGS[tag]) {
+            if (closing) {
+                if ((current == null ? void 0 : current.tag) === tag) flush();
+                continue;
+            }
+            flush();
+            current = heading ? {
+                kind: 'heading',
+                level: Number(heading[1]),
+                tag
+            } : {
+                kind: BLOCK_TAGS[tag],
+                tag
+            };
+            continue;
+        }
+        if (BREAK_TAGS.has(tag)) {
+            if (tag === 'br' && current) {
+                buffer += ' ';
+                continue;
+            }
+            flush();
+        }
+    }
+    flush();
+    return {
+        title,
+        description,
+        lang,
+        canonical,
+        blocks,
+        images,
+        navLinks,
+        links,
+        forms: {
+            kept: forms.kept,
+            removed: forms.removed
+        },
+        loginForm: false
+    };
+}
+
+//# sourceMappingURL=site-import-html.js.map

package/src/lib/site-import/site-import-html.js.map

@@ -0,0 +1 @@
+{"version":3,"sources":["../../../../../../../libs/plugins/ai/src/lib/site-import/site-import-html.ts"],"sourcesContent":["/**\n * @license\n * Copyright 2026 Aglyn LLC\n *\n * Licensed under the Apache License, Version 2.0 (the \"License\");\n * you may not use this file except in compliance with the License.\n * You may obtain a copy of the License at\n *\n *   http://www.apache.org/licenses/LICENSE-2.0\n *\n * Unless required by applicable law or agreed to in writing, software\n * distributed under the License is distributed on an \"AS IS\" BASIS,\n * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.\n * See the License for the specific language governing permissions and\n * limitations under the License.\n */\n\n/*==========================================\n * WHAT A CRAWLED PAGE SAYS (AGL-3974, parent AGL-3973).\n *\n * \"Bring my site\" reads a person's existing pages for their words, their\n * structure and their pictures. No server HTML parser is a dependency of any\n * lib, and a page's text in order is a small question, so this is a\n * tokenizer: tags and text runs, with the few elements that matter tracked\n * and everything else read through.\n *\n * What never survives, before a single tag is walked:\n * - scripts, styles, iframes, embeds, objects, SVG, canvas, media players,\n *   templates and comments, content and all;\n * - any form with a card, bank or payment field, whole;\n * - and a page with a password field is not read at all: it is refused and\n *   reported (`loginForm`), because a sign-in page is exactly what a\n *   phishing clone wants and nothing a site import needs.\n *\n * The output is DATA. Text from a page reaches a model only fenced as\n * untrusted (`fenceUntrustedText`), never as instructions.\n *=========================================*/\n\n/** One run of a page's content, in the order the page draws it. */\nexport type SiteImportBlock =\n  | { kind: 'heading'; level: 1 | 2 | 3 | 4 | 5 | 6; text: string; region: SiteImportRegion }\n  | { kind: 'paragraph' | 'list-item' | 'quote'; text: string; region: SiteImportRegion }\n  | { kind: 'image'; url: string; alt: string; region: SiteImportRegion }\n\n/** Where on the page a block sat: the header and footer carry business details. */\nexport type SiteImportRegion = 'header' | 'main' | 'footer'\n\nexport interface SiteImportImage {\n  /** Absolute https (or http) address. */\n  url: string\n  alt: string\n  /** On the site's own origin; otherwise a CDN or a third party. The importer decides. */\n  sameSite: boolean\n}\n\nexport interface SiteImportNavLink {\n  text: string\n  url: string\n}\n\n/** A form that was kept: what it asks, never what was typed. */\nexport interface SiteImportForm {\n  fields: { type: string; name: string }[]\n}\n\nexport interface SiteImportExtract {\n  title: string\n  description: string\n  lang: string | null\n  canonical: string | null\n  blocks: SiteImportBlock[]\n  images: SiteImportImage[]\n  navLinks: SiteImportNavLink[]\n  /** Same-site page addresses, fragment-free, in first-seen order. */\n  links: string[]\n  forms: { kept: SiteImportForm[]; removed: number }\n  /** True when the page has a password field; nothing else is extracted. */\n  loginForm: boolean\n}\n\n/** Ceilings that keep one odd page from becoming the whole inventory. */\nexport const SITE_IMPORT_PAGE_LIMITS = {\n  blocks: 400,\n  blockChars: 2_000,\n  images: 100,\n  links: 300,\n  navLinks: 60,\n  forms: 10,\n  fieldsPerForm: 30,\n} as const\n\n/** Elements removed with everything inside them. */\nconst REMOVED_WITH_CONTENT = [\n  'script',\n  'style',\n  'noscript',\n  'template',\n  'iframe',\n  'frame',\n  'frameset',\n  'object',\n  'applet',\n  'svg',\n  'math',\n  'canvas',\n  'video',\n  'audio',\n]\n\n/** Form controls, removed with their content once the forms are summarized. */\nconst FORM_CONTROLS = ['textarea', 'select', 'button']\n\n/** Void elements removed wherever they appear. */\nconst REMOVED_VOID = /<(?:embed|param|source|track|link|base)\\b(?:[^>\"']|\"[^\"]*\"|'[^']*')*>/gi\n\nconst PASSWORD_INPUT = /<input\\b(?:[^>\"']|\"[^\"]*\"|'[^']*')*\\btype\\s*=\\s*[\"']?password\\b/i\n\n/**\n * A field that takes money or a bank detail: an autocomplete hint for a\n * card, or a name or id that says card, CVC, IBAN, routing or payment.\n */\nconst PAYMENT_FIELD =\n  /<(?:input|select)\\b(?:[^>\"']|\"[^\"]*\"|'[^']*')*?(?:\\bautocomplete\\s*=\\s*[\"']?\\s*cc-|\\b(?:name|id|data-stripe|data-braintree-name)\\s*=\\s*[\"']?[^\"'\\s>]*(?:card|cc[-_]?(?:num|number|exp|csc|cvv|cvc)|cvv|cvc|csc|iban|routing|account[-_]?number|payment|billing[-_]?card))/i\n\nconst BLOCK_TAGS: Record<string, 'paragraph' | 'list-item' | 'quote'> = {\n  p: 'paragraph',\n  li: 'list-item',\n  dt: 'paragraph',\n  dd: 'paragraph',\n  td: 'paragraph',\n  th: 'paragraph',\n  address: 'paragraph',\n  pre: 'paragraph',\n  figcaption: 'paragraph',\n  caption: 'paragraph',\n  label: 'paragraph',\n  blockquote: 'quote',\n}\n\n/** Tags that end a run of loose text without being a block themselves. */\nconst BREAK_TAGS = new Set([\n  'div',\n  'section',\n  'article',\n  'main',\n  'aside',\n  'header',\n  'footer',\n  'nav',\n  'ul',\n  'ol',\n  'dl',\n  'table',\n  'tr',\n  'tbody',\n  'thead',\n  'figure',\n  'form',\n  'hr',\n  'br',\n  'body',\n])\n\nconst NAMED_ENTITIES: Record<string, string> = {\n  amp: '&',\n  lt: '<',\n  gt: '>',\n  quot: '\"',\n  apos: \"'\",\n  nbsp: ' ',\n  rsquo: '’',\n  lsquo: '‘',\n  rdquo: '”',\n  ldquo: '“',\n  mdash: '—',\n  ndash: '–',\n  hellip: '…',\n  copy: '©',\n  reg: '®',\n  trade: '™',\n  middot: '·',\n  bull: '•',\n  eacute: 'é',\n  egrave: 'è',\n  aacute: 'á',\n  ntilde: 'ñ',\n  uuml: 'ü',\n  ouml: 'ö',\n  auml: 'ä',\n  deg: '°',\n  times: '×',\n}\n\n/** Character references to characters; an unknown name stays as written. */\nexport function decodeHtmlEntities(text: string): string {\n  return text.replace(/&(#x[0-9a-f]+|#\\d+|[a-z][a-z0-9]*);/gi, (whole, body: string) => {\n    if (body[0] === '#') {\n      const code =\n        body[1] === 'x' || body[1] === 'X' ? parseInt(body.slice(2), 16) : parseInt(body.slice(1), 10)\n      if (!Number.isFinite(code) || code <= 0 || code > 0x10ffff) return ''\n      if (code < 32 && code !== 9 && code !== 10 && code !== 13) return ''\n      return String.fromCodePoint(code)\n    }\n    const named = NAMED_ENTITIES[body.toLowerCase()]\n    return named ?? whole\n  })\n}\n\nconst ATTRIBUTE = /([^\\s=/>\"']+)(?:\\s*=\\s*(?:\"([^\"]*)\"|'([^']*)'|([^\\s>\"']+)))?/g\n\n/** A tag's attributes, names lowercased, values decoded. */\nexport function parseAttributes(source: string): Record<string, string> {\n  const attributes: Record<string, string> = {}\n  for (const match of source.matchAll(ATTRIBUTE)) {\n    const name = match[1].toLowerCase()\n    if (name in attributes) continue\n    attributes[name] = decodeHtmlEntities(match[2] ?? match[3] ?? match[4] ?? '')\n  }\n  return attributes\n}\n\nconst TAG = /<(\\/?)([a-zA-Z][\\w:-]*)((?:[^>\"']|\"[^\"]*\"|'[^']*')*)>/g\n\n/** Every `<meta>` tag's attributes. */\nfunction metaTags(html: string): Record<string, string>[] {\n  const out: Record<string, string>[] = []\n  for (const match of html.matchAll(/<meta\\b((?:[^>\"']|\"[^\"]*\"|'[^']*')*)>/gi)) {\n    out.push(parseAttributes(match[1]))\n  }\n  return out\n}\n\n/** The `content` of every `<meta name=\"…\">` with this name. */\nexport function readMetaValues(html: string, name: string): string[] {\n  const wanted = name.toLowerCase()\n  return metaTags(html)\n    .filter((attributes) => (attributes['name'] ?? '').toLowerCase() === wanted)\n    .map((attributes) => (attributes['content'] ?? '').trim())\n    .filter(Boolean)\n}\n\nfunction collapse(text: string): string {\n  return text.replace(/\\s+/g, ' ').trim()\n}\n\nfunction clip(text: string, max: number): string {\n  return text.length > max ? `${text.slice(0, max - 1).trimEnd()}…` : text\n}\n\nfunction removeElements(html: string, tags: readonly string[]): string {\n  let out = html\n  for (const tag of tags) {\n    const paired = new RegExp(`<${tag}\\\\b(?:[^>\"']|\"[^\"]*\"|'[^']*')*>[\\\\s\\\\S]*?<\\\\/${tag}\\\\s*>`, 'gi')\n    out = out.replace(paired, ' ')\n    // One left open runs to the end of the document, as a browser reads it.\n    const unclosed = new RegExp(`<${tag}\\\\b(?:[^>\"']|\"[^\"]*\"|'[^']*')*>[\\\\s\\\\S]*$`, 'i')\n    out = out.replace(unclosed, ' ')\n  }\n  return out\n}\n\n/** Forms with a payment field go whole; the rest are summarized, then go. */\nfunction takeForms(html: string): { html: string; kept: SiteImportForm[]; removed: number } {\n  const kept: SiteImportForm[] = []\n  let removed = 0\n  const out = html.replace(/<form\\b(?:[^>\"']|\"[^\"]*\"|'[^']*')*>[\\s\\S]*?(?:<\\/form\\s*>|$)/gi, (form) => {\n    if (PAYMENT_FIELD.test(form)) {\n      removed += 1\n      return ' '\n    }\n    if (kept.length < SITE_IMPORT_PAGE_LIMITS.forms) {\n      const fields: SiteImportForm['fields'] = []\n      for (const match of form.matchAll(/<(input|textarea|select)\\b((?:[^>\"']|\"[^\"]*\"|'[^']*')*)>/gi)) {\n        const attributes = parseAttributes(match[2])\n        const type = match[1].toLowerCase() === 'input' ? (attributes['type'] || 'text').toLowerCase() : match[1].toLowerCase()\n        if (type === 'hidden' || type === 'submit' || type === 'button' || type === 'image') continue\n        fields.push({ type, name: clip(attributes['name'] || attributes['id'] || '', 80) })\n        if (fields.length >= SITE_IMPORT_PAGE_LIMITS.fieldsPerForm) break\n      }\n      kept.push({ fields })\n    }\n    return ' '\n  })\n  return { html: out, kept, removed }\n}\n\n/** `raw` against the page, as an absolute http(s) address without a fragment. */\nexport function resolveSiteUrl(raw: string | undefined, base: string): URL | null {\n  const value = String(raw ?? '').trim()\n  if (!value || /^(?:javascript|data|mailto|tel|sms|blob|about):/i.test(value)) return null\n  try {\n    const url = new URL(value, base)\n    if (url.protocol !== 'https:' && url.protocol !== 'http:') return null\n    url.hash = ''\n    return url\n  } catch {\n    return null\n  }\n}\n\n/** The first address a `srcset` offers. */\nfunction firstSrcset(srcset: string | undefined): string | undefined {\n  const first = String(srcset ?? '').split(',')[0]?.trim().split(/\\s+/)[0]\n  return first || undefined\n}\n\n/**\n * The words, structure, pictures and links of one page.\n *\n * `isSameSite(url)` says whether an address is on the site being imported;\n * only those become `links`, and images are marked by it.\n */\nexport function extractSiteImportPage(\n  html: string,\n  pageUrl: string,\n  isSameSite: (url: URL) => boolean,\n): SiteImportExtract {\n  const empty: SiteImportExtract = {\n    title: '',\n    description: '',\n    lang: null,\n    canonical: null,\n    blocks: [],\n    images: [],\n    navLinks: [],\n    links: [],\n    forms: { kept: [], removed: 0 },\n    loginForm: false,\n  }\n  let source = String(html ?? '')\n    .replace(/<!--[\\s\\S]*?(?:-->|$)/g, ' ')\n    .replace(/<!\\[CDATA\\[[\\s\\S]*?\\]\\]>/g, ' ')\n  if (PASSWORD_INPUT.test(source)) return { ...empty, loginForm: true }\n\n  source = removeElements(source, REMOVED_WITH_CONTENT).replace(REMOVED_VOID, ' ')\n  // Forms next, so a field list is read before the controls go.\n  const forms = takeForms(source)\n  source = removeElements(forms.html, FORM_CONTROLS)\n\n  const metas = metaTags(source)\n  const metaContent = (key: string, attribute: 'name' | 'property') =>\n    collapse(metas.find((m) => (m[attribute] ?? '').toLowerCase() === key)?.['content'] ?? '')\n  const titleMatch = /<title\\b[^>]*>([\\s\\S]*?)<\\/title\\s*>/i.exec(source)\n  const title = clip(\n    collapse(decodeHtmlEntities(titleMatch?.[1] ?? '')) || metaContent('og:title', 'property'),\n    300,\n  )\n  const description = clip(\n    metaContent('description', 'name') || metaContent('og:description', 'property'),\n    1_000,\n  )\n  const htmlTag = /<html\\b((?:[^>\"']|\"[^\"]*\"|'[^']*')*)>/i.exec(source)\n  const lang = htmlTag ? parseAttributes(htmlTag[1])['lang']?.trim() || null : null\n  let canonical: string | null = null\n  for (const match of String(html ?? '').matchAll(/<link\\b((?:[^>\"']|\"[^\"]*\"|'[^']*')*)>/gi)) {\n    const attributes = parseAttributes(match[1])\n    if ((attributes['rel'] ?? '').toLowerCase().split(/\\s+/).includes('canonical')) {\n      canonical = resolveSiteUrl(attributes['href'], pageUrl)?.toString() ?? null\n      break\n    }\n  }\n\n  const bodyStart = source.search(/<body\\b/i)\n  const body = bodyStart >= 0 ? source.slice(bodyStart) : source.replace(/<head\\b[\\s\\S]*?<\\/head\\s*>/i, ' ')\n  const hasNav = /<nav\\b/i.test(body)\n\n  const blocks: SiteImportBlock[] = []\n  const images: SiteImportImage[] = []\n  const imageSeen = new Set<string>()\n  const navLinks: SiteImportNavLink[] = []\n  const links: string[] = []\n  const linkSeen = new Set<string>()\n\n  let navDepth = 0\n  let headerDepth = 0\n  let footerDepth = 0\n  let current: { kind: 'heading' | 'paragraph' | 'list-item' | 'quote'; level?: number; tag: string } | null = null\n  let buffer = ''\n  let anchor: { url: URL; text: string } | null = null\n\n  const region = (): SiteImportRegion =>\n    footerDepth > 0 ? 'footer' : headerDepth > 0 ? 'header' : 'main'\n  // A site without a <nav> keeps its menu in the header; its links are the nav.\n  const inNav = () => navDepth > 0 || (!hasNav && headerDepth > 0)\n\n  const flush = () => {\n    const text = clip(collapse(decodeHtmlEntities(buffer)), SITE_IMPORT_PAGE_LIMITS.blockChars)\n    buffer = ''\n    const kind = current?.kind ?? 'paragraph'\n    const level = current?.level\n    current = null\n    if (!text || navDepth > 0 || blocks.length >= SITE_IMPORT_PAGE_LIMITS.blocks) return\n    const last = blocks[blocks.length - 1]\n    if (last && last.kind !== 'image' && last.text === text) return\n    if (kind === 'heading') {\n      blocks.push({ kind, level: (level ?? 2) as 1 | 2 | 3 | 4 | 5 | 6, text, region: region() })\n    } else {\n      blocks.push({ kind, text, region: region() })\n    }\n  }\n\n  for (const match of body.matchAll(new RegExp(`${TAG.source}|([^<]+)|<`, 'g'))) {\n    const text = match[4]\n    if (text !== undefined || match[2] === undefined) {\n      const run = text ?? '<'\n      buffer += run\n      if (anchor) anchor.text += run\n      continue\n    }\n    const closing = match[1] === '/'\n    const tag = match[2].toLowerCase()\n    const attributes = closing ? {} : parseAttributes(match[3] ?? '')\n\n    if (tag === 'a') {\n      if (closing) {\n        if (anchor) {\n          const label = clip(collapse(decodeHtmlEntities(anchor.text)), 120)\n          if (inNav() && label && navLinks.length < SITE_IMPORT_PAGE_LIMITS.navLinks) {\n            navLinks.push({ text: label, url: anchor.url.toString() })\n          }\n          anchor = null\n        }\n        continue\n      }\n      const url = resolveSiteUrl(attributes['href'], pageUrl)\n      anchor = url ? { url, text: '' } : null\n      if (url && isSameSite(url) && links.length < SITE_IMPORT_PAGE_LIMITS.links) {\n        const key = url.toString()\n        if (!linkSeen.has(key)) {\n          linkSeen.add(key)\n          links.push(key)\n        }\n      }\n      continue\n    }\n\n    if (tag === 'img' && !closing) {\n      const url = resolveSiteUrl(\n        attributes['src'] || attributes['data-src'] || firstSrcset(attributes['srcset'] || attributes['data-srcset']),\n        pageUrl,\n      )\n      if (url && images.length < SITE_IMPORT_PAGE_LIMITS.images) {\n        const key = url.toString()\n        const alt = clip(collapse(attributes['alt'] ?? ''), 300)\n        if (!imageSeen.has(key)) {\n          imageSeen.add(key)\n          images.push({ url: key, alt, sameSite: isSameSite(url) })\n          if (navDepth === 0) {\n            flush()\n            if (blocks.length < SITE_IMPORT_PAGE_LIMITS.blocks) {\n              blocks.push({ kind: 'image', url: key, alt, region: region() })\n            }\n          }\n        }\n      }\n      continue\n    }\n\n    if (tag === 'nav' || tag === 'header' || tag === 'footer') {\n      flush()\n      const step = closing ? -1 : 1\n      if (tag === 'nav') navDepth = Math.max(0, navDepth + step)\n      if (tag === 'header') headerDepth = Math.max(0, headerDepth + step)\n      if (tag === 'footer') footerDepth = Math.max(0, footerDepth + step)\n      continue\n    }\n\n    const heading = /^h([1-6])$/.exec(tag)\n    if (heading || BLOCK_TAGS[tag]) {\n      if (closing) {\n        if (current?.tag === tag) flush()\n        continue\n      }\n      flush()\n      current = heading\n        ? { kind: 'heading', level: Number(heading[1]), tag }\n        : { kind: BLOCK_TAGS[tag], tag }\n      continue\n    }\n\n    if (BREAK_TAGS.has(tag)) {\n      if (tag === 'br' && current) {\n        buffer += ' '\n        continue\n      }\n      flush()\n    }\n  }\n  flush()\n\n  return {\n    title,\n    description,\n    lang,\n    canonical,\n    blocks,\n    images,\n    navLinks,\n    links,\n    forms: { kept: forms.kept, removed: forms.removed },\n    loginForm: false,\n  }\n}\n"],"names":["SITE_IMPORT_PAGE_LIMITS","blocks","blockChars","images","links","navLinks","forms","fieldsPerForm","REMOVED_WITH_CONTENT","FORM_CONTROLS","REMOVED_VOID","PASSWORD_INPUT","PAYMENT_FIELD","BLOCK_TAGS","p","li","dt","dd","td","th","address","pre","figcaption","caption","label","blockquote","BREAK_TAGS","Set","NAMED_ENTITIES","amp","lt","gt","quot","apos","nbsp","rsquo","lsquo","rdquo","ldquo","mdash","ndash","hellip","copy","reg","trade","middot","bull","eacute","egrave","aacute","ntilde","uuml","ouml","auml","deg","times","decodeHtmlEntities","text","replace","whole","body","code","parseInt","slice","Number","isFinite","String","fromCodePoint","named","toLowerCase","ATTRIBUTE","parseAttributes","source","attributes","match","matchAll","name","TAG","metaTags","html","out","push","readMetaValues","wanted","filter","map","trim","Boolean","collapse","clip","max","length","trimEnd","removeElements","tags","tag","paired","RegExp","unclosed","takeForms","kept","removed","form","test","fields","type","resolveSiteUrl","raw","base","value","url","URL","protocol","hash","firstSrcset","srcset","first","split","undefined","extractSiteImportPage","pageUrl","isSameSite","empty","title","description","lang","canonical","loginForm","metas","metaContent","key","attribute","find","m","titleMatch","exec","htmlTag","includes","toString","bodyStart","search","hasNav","imageSeen","linkSeen","navDepth","headerDepth","footerDepth","current","buffer","anchor","region","inNav","flush","kind","level","last","run","closing","has","add","alt","sameSite","step","Math","heading"],"mappings":"AAAA;;;;;;;;;;;;;;;CAeC,GAED;;;;;;;;;;;;;;;;;;;2CAmB2C,GAE3C,iEAAiE;AA0CjE,uEAAuE,GACvE,OAAO,MAAMA,0BAA0B;IACrCC,QAAQ;IACRC,YAAY;IACZC,QAAQ;IACRC,OAAO;IACPC,UAAU;IACVC,OAAO;IACPC,eAAe;AACjB,EAAU;AAEV,kDAAkD,GAClD,MAAMC,uBAAuB;IAC3B;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;CACD;AAED,6EAA6E,GAC7E,MAAMC,gBAAgB;IAAC;IAAY;IAAU;CAAS;AAEtD,gDAAgD,GAChD,MAAMC,eAAe;AAErB,MAAMC,iBAAiB;AAEvB;;;CAGC,GACD,MAAMC,gBACJ;AAEF,MAAMC,aAAkE;IACtEC,GAAG;IACHC,IAAI;IACJC,IAAI;IACJC,IAAI;IACJC,IAAI;IACJC,IAAI;IACJC,SAAS;IACTC,KAAK;IACLC,YAAY;IACZC,SAAS;IACTC,OAAO;IACPC,YAAY;AACd;AAEA,wEAAwE,GACxE,MAAMC,aAAa,IAAIC,IAAI;IACzB;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;IACA;CACD;AAED,MAAMC,iBAAyC;IAC7CC,KAAK;IACLC,IAAI;IACJC,IAAI;IACJC,MAAM;IACNC,MAAM;IACNC,MAAM;IACNC,OAAO;IACPC,OAAO;IACPC,OAAO;IACPC,OAAO;IACPC,OAAO;IACPC,OAAO;IACPC,QAAQ;IACRC,MAAM;IACNC,KAAK;IACLC,OAAO;IACPC,QAAQ;IACRC,MAAM;IACNC,QAAQ;IACRC,QAAQ;IACRC,QAAQ;IACRC,QAAQ;IACRC,MAAM;IACNC,MAAM;IACNC,MAAM;IACNC,KAAK;IACLC,OAAO;AACT;AAEA,0EAA0E,GAC1E,OAAO,SAASC,mBAAmBC,IAAY;IAC7C,OAAOA,KAAKC,OAAO,CAAC,yCAAyC,CAACC,OAAOC;QACnE,IAAIA,IAAI,CAAC,EAAE,KAAK,KAAK;YACnB,MAAMC,OACJD,IAAI,CAAC,EAAE,KAAK,OAAOA,IAAI,CAAC,EAAE,KAAK,MAAME,SAASF,KAAKG,KAAK,CAAC,IAAI,MAAMD,SAASF,KAAKG,KAAK,CAAC,IAAI;YAC7F,IAAI,CAACC,OAAOC,QAAQ,CAACJ,SAASA,QAAQ,KAAKA,OAAO,UAAU,OAAO;YACnE,IAAIA,OAAO,MAAMA,SAAS,KAAKA,SAAS,MAAMA,SAAS,IAAI,OAAO;YAClE,OAAOK,OAAOC,aAAa,CAACN;QAC9B;QACA,MAAMO,QAAQxC,cAAc,CAACgC,KAAKS,WAAW,GAAG;QAChD,OAAOD,gBAAAA,QAAST;IAClB;AACF;AAEA,MAAMW,YAAY;AAElB,0DAA0D,GAC1D,OAAO,SAASC,gBAAgBC,MAAc;IAC5C,MAAMC,aAAqC,CAAC;IAC5C,KAAK,MAAMC,SAASF,OAAOG,QAAQ,CAACL,WAAY;YAGRI,MAAAA,OAAAA;QAFtC,MAAME,OAAOF,KAAK,CAAC,EAAE,CAACL,WAAW;QACjC,IAAIO,QAAQH,YAAY;QACxBA,UAAU,CAACG,KAAK,GAAGpB,oBAAmBkB,QAAAA,SAAAA,UAAAA,KAAK,CAAC,EAAE,YAARA,UAAYA,KAAK,CAAC,EAAE,YAApBA,QAAwBA,KAAK,CAAC,EAAE,YAAhCA,OAAoC;IAC5E;IACA,OAAOD;AACT;AAEA,MAAMI,MAAM;AAEZ,qCAAqC,GACrC,SAASC,SAASC,IAAY;IAC5B,MAAMC,MAAgC,EAAE;IACxC,KAAK,MAAMN,SAASK,KAAKJ,QAAQ,CAAC,2CAA4C;QAC5EK,IAAIC,IAAI,CAACV,gBAAgBG,KAAK,CAAC,EAAE;IACnC;IACA,OAAOM;AACT;AAEA,6DAA6D,GAC7D,OAAO,SAASE,eAAeH,IAAY,EAAEH,IAAY;IACvD,MAAMO,SAASP,KAAKP,WAAW;IAC/B,OAAOS,SAASC,MACbK,MAAM,CAAC,CAACX;YAAgBA;eAAD,EAACA,mBAAAA,UAAU,CAAC,OAAO,YAAlBA,mBAAsB,IAAIJ,WAAW,OAAOc;OACpEE,GAAG,CAAC,CAACZ;YAAgBA;eAAD,EAACA,sBAAAA,UAAU,CAAC,UAAU,YAArBA,sBAAyB,IAAIa,IAAI;OACtDF,MAAM,CAACG;AACZ;AAEA,SAASC,SAAS/B,IAAY;IAC5B,OAAOA,KAAKC,OAAO,CAAC,QAAQ,KAAK4B,IAAI;AACvC;AAEA,SAASG,KAAKhC,IAAY,EAAEiC,GAAW;IACrC,OAAOjC,KAAKkC,MAAM,GAAGD,MAAM,GAAGjC,KAAKM,KAAK,CAAC,GAAG2B,MAAM,GAAGE,OAAO,GAAG,CAAC,CAAC,GAAGnC;AACtE;AAEA,SAASoC,eAAed,IAAY,EAAEe,IAAuB;IAC3D,IAAId,MAAMD;IACV,KAAK,MAAMgB,OAAOD,KAAM;QACtB,MAAME,SAAS,IAAIC,OAAO,CAAC,CAAC,EAAEF,IAAI,6CAA6C,EAAEA,IAAI,KAAK,CAAC,EAAE;QAC7Ff,MAAMA,IAAItB,OAAO,CAACsC,QAAQ;QAC1B,wEAAwE;QACxE,MAAME,WAAW,IAAID,OAAO,CAAC,CAAC,EAAEF,IAAI,yCAAyC,CAAC,EAAE;QAChFf,MAAMA,IAAItB,OAAO,CAACwC,UAAU;IAC9B;IACA,OAAOlB;AACT;AAEA,2EAA2E,GAC3E,SAASmB,UAAUpB,IAAY;IAC7B,MAAMqB,OAAyB,EAAE;IACjC,IAAIC,UAAU;IACd,MAAMrB,MAAMD,KAAKrB,OAAO,CAAC,kEAAkE,CAAC4C;QAC1F,IAAI1F,cAAc2F,IAAI,CAACD,OAAO;YAC5BD,WAAW;YACX,OAAO;QACT;QACA,IAAID,KAAKT,MAAM,GAAG3F,wBAAwBM,KAAK,EAAE;YAC/C,MAAMkG,SAAmC,EAAE;YAC3C,KAAK,MAAM9B,SAAS4B,KAAK3B,QAAQ,CAAC,8DAA+D;gBAC/F,MAAMF,aAAaF,gBAAgBG,KAAK,CAAC,EAAE;gBAC3C,MAAM+B,OAAO/B,KAAK,CAAC,EAAE,CAACL,WAAW,OAAO,UAAU,AAACI,CAAAA,UAAU,CAAC,OAAO,IAAI,MAAK,EAAGJ,WAAW,KAAKK,KAAK,CAAC,EAAE,CAACL,WAAW;gBACrH,IAAIoC,SAAS,YAAYA,SAAS,YAAYA,SAAS,YAAYA,SAAS,SAAS;gBACrFD,OAAOvB,IAAI,CAAC;oBAAEwB;oBAAM7B,MAAMa,KAAKhB,UAAU,CAAC,OAAO,IAAIA,UAAU,CAAC,KAAK,IAAI,IAAI;gBAAI;gBACjF,IAAI+B,OAAOb,MAAM,IAAI3F,wBAAwBO,aAAa,EAAE;YAC9D;YACA6F,KAAKnB,IAAI,CAAC;gBAAEuB;YAAO;QACrB;QACA,OAAO;IACT;IACA,OAAO;QAAEzB,MAAMC;QAAKoB;QAAMC;IAAQ;AACpC;AAEA,+EAA+E,GAC/E,OAAO,SAASK,eAAeC,GAAuB,EAAEC,IAAY;IAClE,MAAMC,QAAQ3C,OAAOyC,cAAAA,MAAO,IAAIrB,IAAI;IACpC,IAAI,CAACuB,SAAS,mDAAmDN,IAAI,CAACM,QAAQ,OAAO;IACrF,IAAI;QACF,MAAMC,MAAM,IAAIC,IAAIF,OAAOD;QAC3B,IAAIE,IAAIE,QAAQ,KAAK,YAAYF,IAAIE,QAAQ,KAAK,SAAS,OAAO;QAClEF,IAAIG,IAAI,GAAG;QACX,OAAOH;IACT,EAAE,eAAM;QACN,OAAO;IACT;AACF;AAEA,yCAAyC,GACzC,SAASI,YAAYC,MAA0B;QAC/BjD;IAAd,MAAMkD,SAAQlD,iBAAAA,OAAOiD,iBAAAA,SAAU,IAAIE,KAAK,CAAC,IAAI,CAAC,EAAE,qBAAlCnD,eAAoCoB,IAAI,GAAG+B,KAAK,CAAC,MAAM,CAAC,EAAE;IACxE,OAAOD,SAASE;AAClB;AAEA;;;;;CAKC,GACD,OAAO,SAASC,sBACdxC,IAAY,EACZyC,OAAe,EACfC,UAAiC;;QAqCVlD;IAnCvB,MAAMmD,QAA2B;QAC/BC,OAAO;QACPC,aAAa;QACbC,MAAM;QACNC,WAAW;QACX7H,QAAQ,EAAE;QACVE,QAAQ,EAAE;QACVE,UAAU,EAAE;QACZD,OAAO,EAAE;QACTE,OAAO;YAAE8F,MAAM,EAAE;YAAEC,SAAS;QAAE;QAC9B0B,WAAW;IACb;IACA,IAAIvD,SAASN,OAAOa,eAAAA,OAAQ,IACzBrB,OAAO,CAAC,0BAA0B,KAClCA,OAAO,CAAC,6BAA6B;IACxC,IAAI/C,eAAe4F,IAAI,CAAC/B,SAAS,OAAO,aAAKkD;QAAOK,WAAW;;IAE/DvD,SAASqB,eAAerB,QAAQhE,sBAAsBkD,OAAO,CAAChD,cAAc;IAC5E,8DAA8D;IAC9D,MAAMJ,QAAQ6F,UAAU3B;IACxBA,SAASqB,eAAevF,MAAMyE,IAAI,EAAEtE;IAEpC,MAAMuH,QAAQlD,SAASN;IACvB,MAAMyD,cAAc,CAACC,KAAaC;;YACvBH;eAATxC,kBAASwC,cAAAA,MAAMI,IAAI,CAAC,CAACC;gBAAOA;mBAAD,EAACA,eAAAA,CAAC,CAACF,UAAU,YAAZE,eAAgB,IAAIhE,WAAW,OAAO6D;+BAAzDF,WAA+D,CAAC,UAAU,mBAAI;;IACzF,MAAMM,aAAa,wCAAwCC,IAAI,CAAC/D;IAChE,MAAMmD,QAAQlC,KACZD,SAAShC,2BAAmB8E,8BAAAA,UAAY,CAAC,EAAE,mBAAI,QAAQL,YAAY,YAAY,aAC/E;IAEF,MAAML,cAAcnC,KAClBwC,YAAY,eAAe,WAAWA,YAAY,kBAAkB,aACpE;IAEF,MAAMO,UAAU,yCAAyCD,IAAI,CAAC/D;IAC9D,MAAMqD,OAAOW,UAAUjE,EAAAA,wBAAAA,gBAAgBiE,OAAO,CAAC,EAAE,CAAC,CAAC,OAAO,qBAAnCjE,sBAAqCe,IAAI,OAAM,OAAO;IAC7E,IAAIwC,YAA2B;IAC/B,KAAK,MAAMpD,SAASR,OAAOa,eAAAA,OAAQ,IAAIJ,QAAQ,CAAC,2CAA4C;YAErFF;QADL,MAAMA,aAAaF,gBAAgBG,KAAK,CAAC,EAAE;QAC3C,IAAI,EAACD,kBAAAA,UAAU,CAAC,MAAM,YAAjBA,kBAAqB,IAAIJ,WAAW,GAAGgD,KAAK,CAAC,OAAOoB,QAAQ,CAAC,cAAc;;gBAClE/B;YAAZoB,sBAAYpB,kBAAAA,eAAejC,UAAU,CAAC,OAAO,EAAE+C,6BAAnCd,gBAA6CgC,QAAQ,sBAAM;YACvE;QACF;IACF;IAEA,MAAMC,YAAYnE,OAAOoE,MAAM,CAAC;IAChC,MAAMhF,OAAO+E,aAAa,IAAInE,OAAOT,KAAK,CAAC4E,aAAanE,OAAOd,OAAO,CAAC,+BAA+B;IACtG,MAAMmF,SAAS,UAAUtC,IAAI,CAAC3C;IAE9B,MAAM3D,SAA4B,EAAE;IACpC,MAAME,SAA4B,EAAE;IACpC,MAAM2I,YAAY,IAAInH;IACtB,MAAMtB,WAAgC,EAAE;IACxC,MAAMD,QAAkB,EAAE;IAC1B,MAAM2I,WAAW,IAAIpH;IAErB,IAAIqH,WAAW;IACf,IAAIC,cAAc;IAClB,IAAIC,cAAc;IAClB,IAAIC,UAAyG;IAC7G,IAAIC,SAAS;IACb,IAAIC,SAA4C;IAEhD,MAAMC,SAAS,IACbJ,cAAc,IAAI,WAAWD,cAAc,IAAI,WAAW;IAC5D,8EAA8E;IAC9E,MAAMM,QAAQ,IAAMP,WAAW,KAAM,CAACH,UAAUI,cAAc;IAE9D,MAAMO,QAAQ;;QACZ,MAAM/F,OAAOgC,KAAKD,SAAShC,mBAAmB4F,UAAUpJ,wBAAwBE,UAAU;QAC1FkJ,SAAS;QACT,MAAMK,eAAON,2BAAAA,QAASM,IAAI,mBAAI;QAC9B,MAAMC,QAAQP,2BAAAA,QAASO,KAAK;QAC5BP,UAAU;QACV,IAAI,CAAC1F,QAAQuF,WAAW,KAAK/I,OAAO0F,MAAM,IAAI3F,wBAAwBC,MAAM,EAAE;QAC9E,MAAM0J,OAAO1J,MAAM,CAACA,OAAO0F,MAAM,GAAG,EAAE;QACtC,IAAIgE,QAAQA,KAAKF,IAAI,KAAK,WAAWE,KAAKlG,IAAI,KAAKA,MAAM;QACzD,IAAIgG,SAAS,WAAW;YACtBxJ,OAAOgF,IAAI,CAAC;gBAAEwE;gBAAMC,KAAK,EAAGA,gBAAAA,QAAS;gBAA6BjG;gBAAM6F,QAAQA;YAAS;QAC3F,OAAO;YACLrJ,OAAOgF,IAAI,CAAC;gBAAEwE;gBAAMhG;gBAAM6F,QAAQA;YAAS;QAC7C;IACF;IAEA,KAAK,MAAM5E,SAASd,KAAKe,QAAQ,CAAC,IAAIsB,OAAO,GAAGpB,IAAIL,MAAM,CAAC,UAAU,CAAC,EAAE,MAAO;YAU3BE;QATlD,MAAMjB,OAAOiB,KAAK,CAAC,EAAE;QACrB,IAAIjB,SAAS6D,aAAa5C,KAAK,CAAC,EAAE,KAAK4C,WAAW;YAChD,MAAMsC,MAAMnG,eAAAA,OAAQ;YACpB2F,UAAUQ;YACV,IAAIP,QAAQA,OAAO5F,IAAI,IAAImG;YAC3B;QACF;QACA,MAAMC,UAAUnF,KAAK,CAAC,EAAE,KAAK;QAC7B,MAAMqB,MAAMrB,KAAK,CAAC,EAAE,CAACL,WAAW;QAChC,MAAMI,aAAaoF,UAAU,CAAC,IAAItF,iBAAgBG,UAAAA,KAAK,CAAC,EAAE,YAARA,UAAY;QAE9D,IAAIqB,QAAQ,KAAK;YACf,IAAI8D,SAAS;gBACX,IAAIR,QAAQ;oBACV,MAAM7H,QAAQiE,KAAKD,SAAShC,mBAAmB6F,OAAO5F,IAAI,IAAI;oBAC9D,IAAI8F,WAAW/H,SAASnB,SAASsF,MAAM,GAAG3F,wBAAwBK,QAAQ,EAAE;wBAC1EA,SAAS4E,IAAI,CAAC;4BAAExB,MAAMjC;4BAAOsF,KAAKuC,OAAOvC,GAAG,CAAC4B,QAAQ;wBAAG;oBAC1D;oBACAW,SAAS;gBACX;gBACA;YACF;YACA,MAAMvC,MAAMJ,eAAejC,UAAU,CAAC,OAAO,EAAE+C;YAC/C6B,SAASvC,MAAM;gBAAEA;gBAAKrD,MAAM;YAAG,IAAI;YACnC,IAAIqD,OAAOW,WAAWX,QAAQ1G,MAAMuF,MAAM,GAAG3F,wBAAwBI,KAAK,EAAE;gBAC1E,MAAM8H,MAAMpB,IAAI4B,QAAQ;gBACxB,IAAI,CAACK,SAASe,GAAG,CAAC5B,MAAM;oBACtBa,SAASgB,GAAG,CAAC7B;oBACb9H,MAAM6E,IAAI,CAACiD;gBACb;YACF;YACA;QACF;QAEA,IAAInC,QAAQ,SAAS,CAAC8D,SAAS;YAC7B,MAAM/C,MAAMJ,eACVjC,UAAU,CAAC,MAAM,IAAIA,UAAU,CAAC,WAAW,IAAIyC,YAAYzC,UAAU,CAAC,SAAS,IAAIA,UAAU,CAAC,cAAc,GAC5G+C;YAEF,IAAIV,OAAO3G,OAAOwF,MAAM,GAAG3F,wBAAwBG,MAAM,EAAE;oBAE/BsE;gBAD1B,MAAMyD,MAAMpB,IAAI4B,QAAQ;gBACxB,MAAMsB,MAAMvE,KAAKD,UAASf,kBAAAA,UAAU,CAAC,MAAM,YAAjBA,kBAAqB,KAAK;gBACpD,IAAI,CAACqE,UAAUgB,GAAG,CAAC5B,MAAM;oBACvBY,UAAUiB,GAAG,CAAC7B;oBACd/H,OAAO8E,IAAI,CAAC;wBAAE6B,KAAKoB;wBAAK8B;wBAAKC,UAAUxC,WAAWX;oBAAK;oBACvD,IAAIkC,aAAa,GAAG;wBAClBQ;wBACA,IAAIvJ,OAAO0F,MAAM,GAAG3F,wBAAwBC,MAAM,EAAE;4BAClDA,OAAOgF,IAAI,CAAC;gCAAEwE,MAAM;gCAAS3C,KAAKoB;gCAAK8B;gCAAKV,QAAQA;4BAAS;wBAC/D;oBACF;gBACF;YACF;YACA;QACF;QAEA,IAAIvD,QAAQ,SAASA,QAAQ,YAAYA,QAAQ,UAAU;YACzDyD;YACA,MAAMU,OAAOL,UAAU,CAAC,IAAI;YAC5B,IAAI9D,QAAQ,OAAOiD,WAAWmB,KAAKzE,GAAG,CAAC,GAAGsD,WAAWkB;YACrD,IAAInE,QAAQ,UAAUkD,cAAckB,KAAKzE,GAAG,CAAC,GAAGuD,cAAciB;YAC9D,IAAInE,QAAQ,UAAUmD,cAAciB,KAAKzE,GAAG,CAAC,GAAGwD,cAAcgB;YAC9D;QACF;QAEA,MAAME,UAAU,aAAa7B,IAAI,CAACxC;QAClC,IAAIqE,WAAWvJ,UAAU,CAACkF,IAAI,EAAE;YAC9B,IAAI8D,SAAS;gBACX,IAAIV,CAAAA,2BAAAA,QAASpD,GAAG,MAAKA,KAAKyD;gBAC1B;YACF;YACAA;YACAL,UAAUiB,UACN;gBAAEX,MAAM;gBAAWC,OAAO1F,OAAOoG,OAAO,CAAC,EAAE;gBAAGrE;YAAI,IAClD;gBAAE0D,MAAM5I,UAAU,CAACkF,IAAI;gBAAEA;YAAI;YACjC;QACF;QAEA,IAAIrE,WAAWoI,GAAG,CAAC/D,MAAM;YACvB,IAAIA,QAAQ,QAAQoD,SAAS;gBAC3BC,UAAU;gBACV;YACF;YACAI;QACF;IACF;IACAA;IAEA,OAAO;QACL7B;QACAC;QACAC;QACAC;QACA7H;QACAE;QACAE;QACAD;QACAE,OAAO;YAAE8F,MAAM9F,MAAM8F,IAAI;YAAEC,SAAS/F,MAAM+F,OAAO;QAAC;QAClD0B,WAAW;IACb;AACF"}

package/src/lib/site-import/site-import-images.d.ts

@@ -0,0 +1,54 @@
+/**
+ * @license
+ * Copyright 2026 Aglyn LLC
+ *
+ * Licensed under the Apache License, Version 2.0 (the "License");
+ * you may not use this file except in compliance with the License.
+ * You may obtain a copy of the License at
+ *
+ *   http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+import type { PluginMediaIngest } from '@aglyn/aglyn/plugin-manager/plugin-media-ingest';
+import { type PublicFetchOptions, type PublicFetchResult } from '../jobs/ai-theme-brand-inputs';
+import type { AiSiteImportPage } from '../model/ai-site-import';
+/** The largest picture brought in. */
+export declare const SITE_IMPORT_IMAGE_MAX_BYTES: number;
+/** A picture smaller than this is a spacer or a tracking pixel, not a photo. */
+export declare const SITE_IMPORT_IMAGE_MIN_BYTES = 1024;
+/** One picture's own clock. */
+export declare const SITE_IMPORT_IMAGE_TIMEOUT_MS = 8000;
+/** A picture's address as it is fetched and keyed: https, no fragment or tracking; `null` for one that is not a photo. */
+export declare function siteImportImageUrl(raw: string): string | null;
+/** The key a site's library holds an imported picture by. */
+export declare function siteImportImageKey(url: string): string;
+/** A file name for the library, from the picture's address and type. */
+export declare function siteImportImageFileName(url: string, contentType: string): string;
+export interface SiteImportPhotosInput {
+    hostId: string;
+    /** The member the photos are stored as: the job's creator. */
+    uid: string;
+    domain: string;
+    images: ReadonlyArray<{
+        url: string;
+        alt: string;
+    }>;
+    signal?: AbortSignal;
+    /** When the pass must stop asking for more. */
+    deadlineAt: number;
+}
+export interface SiteImportPhotosDeps {
+    ingest: PluginMediaIngest | null;
+    fetch?: (url: string, options: PublicFetchOptions) => Promise<PublicFetchResult>;
+    now?: () => number;
+}
+/**
+ * Brings one page's pictures into the site's library, reusing any the
+ * library already holds by key. Never throws for anything a picture does.
+ */
+export declare function bringInSiteImportPhotos(input: SiteImportPhotosInput, deps: SiteImportPhotosDeps): Promise<NonNullable<AiSiteImportPage['photos']>>;